Paving Machines: Politics and the Provision of Public Infrastructure in American Cities during the Progressive Era, 1900-1910

Rebecca Menes
Department of Political Science
4289 Bunche Hall Box 951472
Los Angeles CA 90095-1472
Telephone: 310-825-6629 Fax: 310-825-0778

In this paper, part of a larger study of the role of machine politics in the economic development of American cities at the turn of the 20th century, I use a game-theoretic model of patronage politics and a cross sectional analysis of politics and city government performance to investigate how the ability to buy votes relaxes the constraints on politicians to keep taxes down, to provide public goods, and to refrain from stealing government monies. Many democratic governments, in the "emerging" democracies and in the industrialized West, suffer from endemic mis-government and corruption. In democracies, corruption is often embedded in a patronage based political organization - in American political parlance a "political machine". A political machine is apparently able to insulate corrupt elected officials from voter ire by buying votes with favors and jobs, making individual deals with a winning coalition of voters.

American city government at the turn of the 20th century provides an unparalleled opportunity to study how buying votes relaxes the constraints that voters otherwise impose on their government. In 1900 there were roughly 100 cities large enough to develop full-fledged political machines. The cities were small, open economies. The cities were politically and fiscally independent, raising their own revenues from taxes and bonds. City governments were expanding into the production of new public goods, such as sewers, water systems and durably paved streets, whose production could be measured. And finally city-level data survives. The US Bureau of the Census collected financial records and measures of the output of real public goods. Over the following century, political scientists and historians wrote the political histories of individual cities, from which it is possible to develop a subjective coding of city governments as dominated, or not dominated, by machine politics.

A game theoretic public choice model of machine politics produces testable hypotheses concerning the impact of the political machine. The key intuition of the model is that a kleptocrat who buys his way into office will be motivated to provide optimal levels of government goods and services if he can capture the resulting locational rents in higher taxes or graft. The political machine's ability to buy votes relaxes the electoral constraints on the government, allowing the boss to raise taxes and divert funds, but the open city borders force the government to provide the socially optimal quantity of public goods. Public goods are the key to capturing locational rents, however, only if the city boundaries are open to the free movement of people and inputs. If the government does not have to provide public goods to lure and keep residents, then buying the election with private favors allows the kleptocrat to win office while providing less than the optimal quantity of public goods.

I test the fiscal implications of the model against the wages of unskilled city government employees, a proxy for patronage, and the overall municipal budget. The presence of a machine appears to increase municipal spending and municipal wages, consistent with the prediction that the machine boss spends city resources hiring patronage employees in exchange for political support and pays to provide public goods and services. Regression analysis found an 8 percent municipal premium on unskilled wages of city workers and an 18 percent premium in city budgets in machine cities.

The most difficult aspect of government performance to capture empirically is the output of public goods. In general, fiscal data on governments is more available than data on real output. However, the provision of durable paving provides a serendipitous natural test of the ability of turn of the century city government to respond to citizen demands for public goods. Technological innovations in the 1890s in asphalt and concrete meant that cities had access to new, better, lower-cost alternatives for durable paving. Changes in transportation technology increased demand for smooth, durable paving. I can compare paving across cities because the technology for installing paving, and the demand for paving, were relatively more uniform across cities than the supply and demand for most other public goods. And the data survive, bi-annual surveys of the mileage and acreage of streets, paved and unpaved, for the first decade of the century. Therefore I use the proportion of streets with durable paving as a measure of the success of the city in providing public goods. As predicted, holding constant for city physical size, population, ethnic structure, wages and region, the cities known to have machine-dominated governments provide as much or more paving than other cities.

The pattern of spending and of public goods provision in cities in which political machines are observed is consistent with the predictions of the model: 1) municipal machines spend money on patronage hires to buy votes, as indicated by high wages for unskilled municipal workers; 2) municipal machines provide the optimal level of public goods, given the technology and understood best practice of the period, as shown by the provision of paving; and 3) municipal machines inflate the overall budget as a result of providing the full basket of public goods, covering patronage costs of buying votes, and, presumably, fattening private bank accounts through inflated contracts and kickbacks, as suggested by the premium on the per capita budget in machine-dominated cities.

The results suggest that patronage and corruption did exist, and did inflate city budgets, but that the power of a near and open boundary in constrained rent-seeking politicians to provide public goods which fostered the economic development of American cities.